feat: complete commercial developer workflowspublish-commercial-control-plane

Add tenant-safe usage observability, prepaid billing controls, API key lifecycle management, Embeddings metering, configurable billing alerts, and resilient provider health propagation. Harden Stripe failure handling, migrations, readiness, and the authenticated control-plane UI with end-to-end verification evidence.

1 files changed, 30 insertions, 4 deletions

diff --git a/cmd/aigw/main.go b/cmd/aigw/main.go
index 3ac388a..60956a8 100644
--- a/cmd/aigw/main.go
+++ b/cmd/aigw/main.go
@@ -9,6 +9,7 @@ import (
 	"net/http"
 	"os"
 	"os/signal"
+	"strconv"
 	"syscall"
 	"time"
 
@@ -180,6 +181,27 @@ func run(ctx context.Context, cfg config.Config, logger *slog.Logger) error {
 		billingMeter = billingService
 	}
 	routeHealth := providerhealth.New(providerhealth.Options{})
+	var sharedHealth *providerhealth.RedisHistory
+	if cfg.ProviderHealth.SharedHistoryEnabled {
+		instanceID := cfg.Server.DeploymentRegion + ":" + strconv.Itoa(os.Getpid()) + ":" + strconv.FormatInt(time.Now().UnixNano(), 36)
+		sharedHealth = providerhealth.NewRedisHistory(providerhealth.HistoryOptions{
+			Enabled: true, RedisURL: cfg.ControlPlane.RedisURL, Stream: cfg.ProviderHealth.SharedHistoryStream,
+			Instance: instanceID, MaxEvents: cfg.ProviderHealth.SharedHistoryMaxEvents,
+			TTL: time.Duration(cfg.ProviderHealth.SharedHistoryTTLSeconds) * time.Second, Logger: logger, Metrics: metrics,
+		})
+		if sharedHealth != nil {
+			routeHealth.SetSink(sharedHealth)
+			go sharedHealth.Run(ctx, routeHealth)
+			defer sharedHealth.Close()
+		} else {
+			logger.Warn("provider_health_shared_history_disabled", "reason", "redis_not_configured", "fallback", "local")
+		}
+	}
+	go providerhealth.NewProber(providerhealth.ProbeOptions{
+		Enabled: cfg.ProviderHealth.ActiveProbesEnabled, Interval: time.Duration(cfg.ProviderHealth.ProbeIntervalSeconds) * time.Second,
+		Timeout: time.Duration(cfg.ProviderHealth.ProbeTimeoutSeconds) * time.Second, Catalog: modelCatalog,
+		Tracker: routeHealth, Metrics: metrics, Logger: logger,
+	}).Run(ctx)
 
 	inferenceAPI := httpapi.New(httpapi.Options{
 		Authenticator:    authenticator,
@@ -204,10 +226,14 @@ func run(ctx context.Context, cfg config.Config, logger *slog.Logger) error {
 			Logger: logger, Prefix: cfg.Admin.BasePath, RegistrationEnabled: cfg.Admin.RegistrationEnabled,
 			SessionTTL: time.Duration(cfg.Admin.SessionTTLHours) * time.Hour, Currency: cfg.Billing.Currency,
 			PublicURL: cfg.Admin.PublicURL, WebAuthn: webAuthn, MailEnabled: cfg.Admin.Mail.Enabled,
-			InferencePublicURL:      cfg.Admin.InferencePublicURL,
-			DefaultLowBalanceMicros: cfg.Admin.Mail.LowBalanceMicros,
-			Catalog:                 modelCatalog,
-			ProviderHealth:          routeHealth,
+			InferencePublicURL: cfg.Admin.InferencePublicURL,
+			BillingPreferenceDefaults: controlplane.BillingPreferenceDefaults{
+				LowBalanceThresholdMicros: cfg.Admin.Mail.LowBalanceMicros,
+				SpendAnomalyMultiplier:    cfg.Admin.Mail.SpendAnomalyMultiplier,
+				SpendAnomalyMinMicros:     cfg.Admin.Mail.SpendAnomalyMinMicros,
+			},
+			Catalog:        modelCatalog,
+			ProviderHealth: routeHealth,
 		}).Handler()
 	}
 	operationHandler := operations.Handler{Store: store, Manager: manager, Billing: billingService, Metrics: metrics,